Medical Image Pixel Highlighting Using Regional Intensity Distributions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing techniques for identifying lesions in black-and-white contrast organ images, such as brain images, require manual intervention and have low determination accuracy, typically ranging from 60% to 80%, which is not sufficient for precise lesion specification.
Innovation Solution
An image processing apparatus and method that includes a region setting unit, intensity value frequency distribution calculation, difference value calculation, pixel selection, and output processing to highlight pixels based on the difference in intensity value distributions between a region of interest and its vicinity, using a computer system with specific units for image acquisition, region setting, and display processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual marking is used for lesion specification, then diagnostic accuracy can be maintained, but work efficiency deteriorates and the process becomes cumbersome
Solution Approach 1:
The system performs automatic lesion specification by having the image processing apparatus itself analyze the medical image and identify lesion regions without requiring manual intervention. The apparatus calculates intensity value frequency distributions, compares them against reference data, and automatically determines lesion locations, enabling the system to serve its own diagnostic function.
Solution Approach 2:
The patent replaces the manual mechanical marking process with an automated image processing system. Instead of requiring a diagnostician to manually mark lesions on images, the system uses computational algorithms to analyze intensity value distributions and automatically specify lesion regions, substituting human manual work with automated processing.
2Productivity
If automatic lesion determination using predetermined thresholds or machine learning is used, then work efficiency is improved, but determination accuracy deteriorates to only 60% to 80%
Solution Approach 1:
The patent changes the key parameter from simple intensity thresholding or basic machine learning features to intensity value frequency distribution analysis. By calculating and comparing the distribution of intensity values across different regions of the image against reference data, the system achieves more accurate lesion determination while maintaining automated processing efficiency.
Solution Approach 2:
The patent transitions from analyzing single intensity values or simple pixel data to analyzing the entire frequency distribution of intensity values. This adds a dimensional aspect by considering the distribution pattern across multiple intensity levels rather than relying on a single threshold or feature, thereby improving determination accuracy.
Data Source
AI summary
An image processing apparatus includes an image acquisition unit acquiring a medical image, a region setting unit setting a peripheral region around an inner region set in the medical image as a region including a lesion, an intensity value ratio distribution calculation unit calculating a histogram comprising a distribution of intensity value ratios for the inner region and calculating a histogram being a distribution of intensity value ratios for the peripheral region, a ratio difference calculation unit that calculates a ratio difference comprising a difference between intensity value ratios in the inner region and peripheral regions for each of predetermined intensity values, an intensity value determination unit selecting a pixel to be highlighted in the medical image based on the ratio difference, and a display processing unit outputting the medical image whereby the pixel selected by the pixel selection unit is highlighted in the medical image to a display device.